To read a case exhibit well, first orient yourself (title, axes, units, footnotes), then find the one or two insights that matter, quantify them, and say what they mean for the client's decision. Interviewers aren't testing whether you can describe a chart. They're testing whether you can turn data into a "so what" quickly and accurately.
Why do exhibits trip up strong candidates?
Exhibits concentrate several risks into a few seconds: misreading units, getting lost in detail, and describing everything instead of interpreting something. They are also standard across formats. Interviewer-led cases (McKinsey) often hand you a chart and ask a specific question, BCG's chatbot online case includes exhibit-based questions, and McKinsey's Solve Redrock module is built around data interpretation.
What is a reliable method for any exhibit?
Step 1: Orient for 10–20 seconds
Before you say anything substantive, check:
- Title: what is this supposed to show?
- Axes and units: £ or £m? Percentages or percentage points? Per unit or total?
- Time period and segments: which years, regions or customer groups?
- Footnotes and sources: these often hide the crucial caveat, such as "excludes online sales".
Say it aloud briefly: "This shows quarterly revenue by channel, in £m, for 2023 to 2025."
Step 2: Look for the outliers and the trend
Ask which line is moving differently, which bar is largest or smallest, and where the break in the pattern is. Insight almost always sits in the exception, not in the average.
Step 3: Quantify
Don't say "online grew a lot". Say "online roughly doubled from £40m to £80m while stores were flat at about £200m, so online's share rose from about 17% to about 29%." Rough but specific beats vague.
Step 4: Connect to the question
End with the implication: "So the revenue decline isn't a demand problem overall. It's concentrated in stores, which supports looking at store costs and footprint next."
Step 5: Propose the next step
Strong candidates move the case forward: "I'd like to understand whether store traffic or basket size is driving this. Do we have data on footfall?"
What are the most common chart traps?
Percentages versus percentage points
A margin moving from 10% to 12% is a 2 percentage point increase, or a 20% relative increase. Mixing these up is a classic error.
Truncated axes
A y-axis that starts at 80 instead of 0 can make a small change look dramatic. Read the actual values.
Stacked charts
In stacked bars, only the bottom segment has a common baseline. Compare the middle segments by calculating values, not by eye.
Mix effects
A company's average margin can fall even if every product's margin rises, if sales shift towards lower-margin products. When averages move unexpectedly, check the mix.
Absolute versus relative
A segment growing 50% from a tiny base may matter less than one growing 5% from a huge base. Always sense-check materiality.
How should you handle tables full of numbers?
Tables are harder than charts because nothing is visually highlighted. Scan the row and column totals first, then look for the cells that drive the answer. If asked to calculate, state which cells you'll use before calculating. It's fine to say: "Let me take a moment to look at this table."
What if the exhibit seems irrelevant or confusing?
Sometimes an exhibit is a deliberate distraction or contains more data than you need. You can say: "Most of this table is context. The key line for our question seems to be contribution margin by region." If something is genuinely unclear, ask: "Is this revenue or volume?" Clarifying is a strength. Guessing wrongly is a weakness.
How do you practise exhibit reading?
- Use real charts. Annual reports, Office for National Statistics releases and business newspapers are full of charts. Give yourself 30 seconds to state the one-sentence "so what" for each.
- Time yourself. In live interviews you won't get minutes of silence. Aim to orient in under 20 seconds and have an insight within a minute.
- Practise in case context. Insights only count if you link them to the client's question. Platforms such as caize.ai embed exhibits within a full case, so you practise the hand-off from data to recommendation, not just the reading.
What does a strong exhibit response sound like?
Exhibit: a bar chart showing a ferry operator's passengers by route, 2022–2025, with one route falling sharply.
Strong: "This shows annual passengers by route in thousands. Four of the five routes are broadly stable, but Route C fell from about 900k to about 500k, a drop of over 40%. That one route explains most of the overall decline. Given our question is about restoring profitability, I'd want to know whether Route C lost passengers to a competitor, such as a new bridge or airline, or because of something we changed, like timetable cuts. Do we know what happened on that route?"
FAQ
How long should I spend looking at an exhibit before talking?
Usually 10–30 seconds to orient. Say that you're taking a moment, as silence with context is fine.
Should I describe everything on the chart?
No. Mention the orientation briefly, then focus on the one or two insights relevant to the question.
What if I misread a chart?
Correct yourself openly and continue. Interviewers care more about how you recover than about a perfect first read.
Do online assessments include charts?
Yes. BCG's online case and McKinsey Solve both involve data and exhibit interpretation, although formats change, so check your invitation.
